Deep learning for time series forecasting

Abstract

Time series generally have the characteristics such as high noise, non-linear and chaotic. Due to the characteristics of the time series and the existence of big data, it has been becoming to prefer intelligent methods such as deep learning.The aim of this study is to make estimations of time series using deep learning techniques on financial time series. The originality of study is that the stock prices of Borsa Istanbul-100 forecast using popular three methods about deep learning such as Multilayer Perceptrons, Convolutional Neural Networks and Long Short-Term Memory Networks.